PURPOSE: To quantify the target delineation variability in peripheral early stage lung cancer treated with SBRT and derive corresponding margins. METHODS AND MATERIALS: Sixteen early stage NSCLC GTV's were delineated by 11 radiation oncologists from 4 institutes. A median surface was computed and the delineation variation perpendicular to this surface was measured (local standard deviation=SD). The overall target delineation variability was quantified by the root-mean-square (rms) of the local SD. The required margin was determined by expanding all delineations to encompass the median surface, where after the underlying probability distribution was modeled by a number of uncorrelated 'pimples-and-dimples'. RESULTS: The overall target delineation variability was 2.1mm (rms). Institute I-III delineated significantly smaller volumes than institute IV, yielding target delineation variabilities of 1.2mm and 1.8mm respectively. The margin required to obtain 90% coverage of the delineated contours was 3.4mm and 5.9mm respectively. The factor α in M=αΣ required to calculate adequate margins was 2.8-3.2, which is larger than the 2.5 found for 3D rigid target displacement. CONCLUSION: A relatively small target delineation uncertainty of 1.2mm-1.8mm (1SD) was observed for early stage NSCLC. A 3.4-5.9mm GTV-to-PTV margin was required to account for this uncertainty alone, ignoring other sources of geometric uncertainties.